On 1/7/26 3:37 PM, Jakub Kicinski wrote:

On Wed, 7 Jan 2026 11:04:15 -0600 Daniel Jurgens wrote:

quoted

+	if (err && err != -EOPNOTSUPP) {
+		if (netif_running(vi->dev))
+			virtnet_close(vi->dev);
+
+		/* disable_rx_mmode_work takes the rtnl_lock, so just set the
+		 * flag here while holding the lock.
+		 *
+		 * remove_vq_common resets the device and frees the vqs.
+		 */
+		vi->rx_mode_work_enabled = false;
+		disable_delayed_refill(vi);
+		rtnl_unlock();
+		remove_vq_common(vi);
+		return err;

disable_delayed_refill() is going away in net 

https://lore.kernel.org/all/20260106150438.7425-1-minhquangbui99@gmail.com/ (local)

You'll have to wait for that change to propagate to net-next to avoid
a transient build issue:

https://netdev.bots.linux.dev/contest.html?pw-n=0&branch=net-next-2026-01-07--21-00&pw-n=0&pass=0

Thanks for the heads up. The AI bot flagged another bug with the return
value of virtnet_restore_up so I'll send a v16 once this patch comes in.